Re: [Nbd] [ANNOUNCE] Portable NBD server

[Andy Isaacson <adi@...24...>]

On Sat, Jul 17, 2004 at 10:55:36PM +0200, Wouter Verhelst wrote:
> > > Well, you should have fixed the bugs and submit a patch, not
> > > start your own branch...
> > 
> > I agree, if the only reason I were doing this were to serve NBD from
> > non-Linux hosts, then writing a new server would be dumb.  However:
> > 
> > 1. I'm planning to go a radically different direction with mine, so it's
> > unlikely the patches would be (or should be!) accepted.
> 
> You didn't even ask.

It didn't occur to me, sorry.  I've seen too many people go "I have this
great idea, you should do it my way" and then never show up with code; I
didn't want to be one of those.

> I fully agree that nbd-server needs quite a lot of work, and have
> outlined what I plan to do with it[0]. Maybe your ideas aren't even far
> from mine, in which case help (patches, ideas, suggestions, ...) would
> be welcome. Forks aren't, though, because they tend to discourage
> cooperation rather than encourage it. And hey, if you already saw what I
> plan to do but think I'm braindead, tell me about it.
> 
> [0] http://nbd.sf.net/roadmap.html

In general I like the roadmap, and don't think you're braindead.  I want
to prototype some stuff in my own tree before trying to fit it into the
main nbd tree.

I'm sorry that you think I'm promulgating a fork.  I prefer to think of
it as a research prototype; I want to keep my code ASAP (as simple as
possible) and try out a couple of performance ideas.  If they prove to
be useful I don't see any reason not to redo it in the main nbd tree.

Of course the situation isn't the same, but consider Apache versus
thttpd.  There's nothing wrong with having multiple implementations that
speak the same protocol.

-andy